Webbrowser cache implementation.

[I did not see a squid development list, if there is one could someone be
so kind as to forward this there]

Hi.. I'm writing to this list on behalf of Project Mnemonic. Project
Mnemonic is a effort to produce a free and modular web browser for
initally linux, later modern unixes, and much later windows and the
masses. For more information visit http://mnemonic.browser.org/.

[you might want to see the end of this message for a summary of what this
message is about]

In our IRC meeting we discussed some of my ideas about Mnemonic's caching
and I think we might have some ideas that warrent a look over by some
people who know more about internet caching then we do.

I was considering implimenting a sort of pre-caching algrithim where the
browser could pre-load pages off of links. Obviously this raises serious
bandwidth concers so we decided that maby there is reason to define a
internet 'bandwidth conservation protocall' maby as an extension of ICP.

Here is what I see this thing could do:
        
        1) It would define a standard method for the browser to descover
           who the closest proxy is and what sites are closer then it.
        2) It would define a method such that a browser could ask how
           bandwidth agressive it is allow to be.

We have already figured out how we can use ICP when we are talking to a
squid cache so they we can agressivly pre-load hits without causing misses
to be loaded..

Idealy we this protocall could work with ICP or supercede it such that.
The browser could make ICP requests when it is being agressive.. Those
requests would have a flag stating that they are agressive requests and
the cache need no honor them based on a) current traffic b) the type of
link between the site and the computer with the browser.. I.e. agressive
loading of nearby cache misses should never go across intercontinentinal
lines but are ok if they are just one hop across a t1...

Further more we would like to be able to limit cache agressiveness by mime
type.... So that a browser could (if the upstream network okays it) get a
page, pre-dns-resolv all links (should there be under a threshold # of
them, else use something like ICP to only resolv DNS hits from the nearest
proxy), then pre-load the html of all links (providing it's under a size
limit), then pre-load all local cache hits on image mime types...

I'm sorry if this is a bit unclear.. I should have just given a summary:

Heres a summary:
 The Mnemonic Browser Project is looking for some ideas on the
implimentation of an advanced caching system for webbrowsers. We desire
our implimention to be robust, and fast (for a users point of view), while
remaining bandwidth concious. We desire configurable pre-loading/caching
and an integral relationship with internet hierarchial caches. The project
always welcomes additional developers.
        
                                                Thanks for your time,
                                                Gregory Maxwell

----
For further information on project Mnemonic please visit:
http://mnemonic.browser.org